\begin{table}

\caption{\label{tab:judgment_data_cleaning}Number of codings across categories retained at subsequent data cleaning steps by sample.}
\centering
\fontsize{9}{11}\selectfont
\begin{tabular}[t]{>{}l>{\raggedleft\arraybackslash}p{0.47in}>{\raggedleft\arraybackslash}p{0.47in}>{\raggedleft\arraybackslash}p{0.47in}>{\raggedleft\arraybackslash}p{0.4in}>{\raggedleft\arraybackslash}p{0.5in}>{\raggedleft\arraybackslash}p{0.1in}>{\raggedleft\arraybackslash}p{0.3in}}
\toprule
\multicolumn{1}{c}{\textbf{ }} & \multicolumn{5}{c}{\textbf{Coding categories}} & \multicolumn{2}{c}{\textbf{ }} \\
\cmidrule(l{3pt}r{3pt}){2-6}
\textbf{Removal steps} & Cannot answer & No & Yes, general & Yes, specific & Yes, but unsure & \texttt{NA} & \emph{Total}\\
\midrule
\addlinespace[0.3em]
\multicolumn{8}{l}{\textit{\textbf{Sample 1}}}\\
\hspace{1em} & 312 & 12803 & 3131 & 2036 & 1338 & 0 & 19620\\
\hspace{1em}codings of ``low-quality'' coders & 269 & 7751 & 2763 & 1736 & 1185 & 0 & 13704\\
\hspace{1em}judgments made in < 4 seconds & 268 & 7697 & 2754 & 1733 & 1181 & 0 & 13633\\
\hspace{1em}majority-``Cannot answer'' tweets & 222 & 7684 & 2748 & 1726 & 1177 & 0 & 13557\\
\hspace{1em}other ``Cannot answer'' codings & 0 & 7684 & 2748 & 1726 & 1177 & 0 & 13335\\
\addlinespace[0.3em]
\hline
\multicolumn{8}{l}{\textit{\textbf{Sample 2}}}\\
\hspace{1em} & 321 & 6053 & 4770 & 3206 & 2220 & 2 & 16572\\
\hspace{1em}codings of ``low-quality'' coders & 220 & 5114 & 4574 & 3113 & 1661 & 0 & 14682\\
\hspace{1em}judgments made in < 4 seconds & 219 & 5000 & 4469 & 2913 & 1649 & 0 & 14250\\
\hspace{1em}majority-``Cannot answer'' tweets & 188 & 4993 & 4463 & 2906 & 1644 & 0 & 14194\\
\hspace{1em}other ``Cannot answer'' codings & 0 & 4993 & 4463 & 2906 & 1644 & 0 & 14006\\
\bottomrule
\end{tabular}
\end{table}
